Update README.md
Browse files
README.md
CHANGED
|
@@ -1,14 +1,21 @@
|
|
| 1 |
---
|
| 2 |
-
base_model:
|
|
|
|
| 3 |
library_name: transformers
|
| 4 |
tags:
|
| 5 |
- mergekit
|
| 6 |
- merge
|
| 7 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8 |
---
|
| 9 |
-
# QwQ-
|
| 10 |
|
| 11 |
-
|
| 12 |
|
| 13 |
## Merge Details
|
| 14 |
### Merge Method
|
|
@@ -18,7 +25,7 @@ This model was merged using the passthrough merge method.
|
|
| 18 |
### Models Merged
|
| 19 |
|
| 20 |
The following models were included in the merge:
|
| 21 |
-
* /
|
| 22 |
|
| 23 |
### Configuration
|
| 24 |
|
|
@@ -30,25 +37,33 @@ merge_method: passthrough
|
|
| 30 |
slices:
|
| 31 |
- sources:
|
| 32 |
- layer_range: [0, 16]
|
| 33 |
-
model: /
|
| 34 |
- sources:
|
| 35 |
- layer_range: [8, 24]
|
| 36 |
-
model: /
|
| 37 |
- sources:
|
| 38 |
- layer_range: [16, 32]
|
| 39 |
-
model: /
|
| 40 |
- sources:
|
| 41 |
- layer_range: [24, 40]
|
| 42 |
-
model: /
|
| 43 |
- sources:
|
| 44 |
- layer_range: [32, 48]
|
| 45 |
-
model: /
|
| 46 |
- sources:
|
| 47 |
- layer_range: [40, 56]
|
| 48 |
-
model: /
|
| 49 |
- sources:
|
| 50 |
- layer_range: [48, 64]
|
| 51 |
-
model: /
|
| 52 |
|
| 53 |
|
| 54 |
```
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
+
base_model:
|
| 3 |
+
- JackCloudman/QwQ-32B-Preview-jackterated
|
| 4 |
library_name: transformers
|
| 5 |
tags:
|
| 6 |
- mergekit
|
| 7 |
- merge
|
| 8 |
+
- uncensored
|
| 9 |
+
- abliterated
|
| 10 |
+
- chat
|
| 11 |
+
license: apache-2.0
|
| 12 |
+
language:
|
| 13 |
+
- en
|
| 14 |
+
pipeline_tag: text-generation
|
| 15 |
---
|
| 16 |
+
# JackCloudman/QwQ-56B-Ghost
|
| 17 |
|
| 18 |
+
**QwQ-56B-Ghost** is an extended model derived from Qwen-32B-Preview-Jackterated using the _merging passthrough_ technique inspired in models like [miqu-1-120b](https://huggingface.co/wolfram/miqu-1-120b). The result is a **~56B parameter model**
|
| 19 |
|
| 20 |
## Merge Details
|
| 21 |
### Merge Method
|
|
|
|
| 25 |
### Models Merged
|
| 26 |
|
| 27 |
The following models were included in the merge:
|
| 28 |
+
* JackCloudman/QwQ-32B-Preview-jackterated
|
| 29 |
|
| 30 |
### Configuration
|
| 31 |
|
|
|
|
| 37 |
slices:
|
| 38 |
- sources:
|
| 39 |
- layer_range: [0, 16]
|
| 40 |
+
model: JackCloudman/QwQ-32B-Preview-jackterated
|
| 41 |
- sources:
|
| 42 |
- layer_range: [8, 24]
|
| 43 |
+
model: JackCloudman/QwQ-32B-Preview-jackterated
|
| 44 |
- sources:
|
| 45 |
- layer_range: [16, 32]
|
| 46 |
+
model: JackCloudman/QwQ-32B-Preview-jackterated
|
| 47 |
- sources:
|
| 48 |
- layer_range: [24, 40]
|
| 49 |
+
model: JackCloudman/QwQ-32B-Preview-jackterated
|
| 50 |
- sources:
|
| 51 |
- layer_range: [32, 48]
|
| 52 |
+
model: JackCloudman/QwQ-32B-Preview-jackterated
|
| 53 |
- sources:
|
| 54 |
- layer_range: [40, 56]
|
| 55 |
+
model: JackCloudman/QwQ-32B-Preview-jackterated
|
| 56 |
- sources:
|
| 57 |
- layer_range: [48, 64]
|
| 58 |
+
model: JackCloudman/QwQ-32B-Preview-jackterated
|
| 59 |
|
| 60 |
|
| 61 |
```
|
| 62 |
+
|
| 63 |
+
|
| 64 |
+
## Credits
|
| 65 |
+
- Qwen and QwQ-32B-Preview
|
| 66 |
+
- @FailSpy [notebook](https://huggingface.co/failspy/llama-3-70B-Instruct-abliterated/blob/main/ortho_cookbook.ipynb) with abliterate technique
|
| 67 |
+
- [Mergekit](https://github.com/cg123/mergekit)
|
| 68 |
+
- [Archive](https://youtu.be/FzmaF5p96pc?si=RqMjDgEh5J_090Js&t=2015)
|
| 69 |
+
- You :D
|